1
$\begingroup$

Is this a valid test for convergence of $\sum_{n=1}^\infty a_n$ where $a_n$ are all positive? Define:

$\rho_n=a_n/a_{n+1}$

The series converges if $\rho_n>1$ for all n>N

The series diverges if $\rho_n \le 1$ for all n>N

where $N$ is some positive integer. Note this is not the same as the usual ratio test which states that the series converges if $\lim_{n \to \infty}\rho_n>1$ and diverges if $\lim_{n \to \infty}\rho_n<1$ with no conclusion for 1.

I ask this because Kummer's test has been stated as: $\rho_n=D_n a_n/a_{n+1}-D_{n+1}$ where $D_n$ is a positive term series, with convergence for $\rho_n>0$ and divergence for $\rho_n \le 0$ and $D_n$ divergent, for some $n>N$. It has also been stated in the limit form where $\lim_{n \to \infty}\rho_n<0$ and $D_n$ divergent for divergence. Substituting $D_n=1$ into Kummers test gives the above statement (no limits), along with the usual ratio test involving limits.

$\endgroup$
1
$\begingroup$

No. For example, your test predicts that $$ \sum_{n \geq 1} \frac{1}{n^2} $$ diverges.

$\endgroup$
0
$\begingroup$

Ok, thanks, now I see the problem. Kummer's $\rho_n$ is $D_n a_n/a_{n+1}-D_{n+1}$ and for convergence, there must be a $c>0$ such that $\rho_n \ge c$ which is NOT the same as $\rho_n > 0$, which I mistakenly supposed. If $a_n=1/n^2$, then $\rho_n=(1+1/n)^2$ which, although it is greater than zero for all $n\ge 1$, there is no $c>0$ that it is greater than or equal to for any $n$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.